ERR_SSL_WRONG_VERSION_NUMBER when running nextjs in docker cointainer with traefik

[Tassil0]

Summary

Nextjs app running in docker, responds with 500 on every request on every route. The app is using traefik for SSL/reverse proxy, other services running the same traefik config have no issues whatsoever. Next server log only shows ERR_SSL_WRONG_VERSION_NUMBER.

Additional information

Next server log:

Error: fetch failed
    at context.fetch (/app/apps/webapp/.next/standalone/node_modules/.pnpm/next@14.0.2_@babel+core@7.23.2_react-dom@18.2.0_react@18.2.0_sass@1.66.1/node_modules/next/dist/server/web/sandbox/context.js:272:38)
    at Object.middleware_middleware [as handler] (/app/apps/webapp/.next/standalone/apps/webapp/.next/server/src/middleware.js:2253:15)
    at /app/apps/webapp/.next/standalone/apps/webapp/.next/server/src/middleware.js:1426:23
    at AsyncLocalStorage.run (node:async_hooks:346:14)
    at Object.wrap (/app/apps/webapp/.next/standalone/apps/webapp/.next/server/src/middleware.js:1238:24)
    at adapter (/app/apps/webapp/.next/standalone/apps/webapp/.next/server/src/middleware.js:1413:53)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5) {
  cause: [Error: D07B0FDCAAAA0000:error:0A00010B:SSL routines:ssl3_get_record:wrong version number:../deps/openssl/openssl/ssl/record/ssl3_record.c:354:
  ] {
    library: 'SSL routines',
    reason: 'wrong version number',
    code: 'ERR_SSL_WRONG_VERSION_NUMBER'
  }
}

The next app is built with option output: standalone; and run using node server.js command.
We use Next.js v14.

• Docker compose file, that is used to run the app: https://pastebin.com/VZWRAqzX
• traefik.yaml: https://pastebin.com/McXyKzZR
• Dockerfile: https://pastebin.com/60gGrwAY
• note: app is built in CI pipeline, then the build output is copied into the image
• app is using pnpm with turborepo monorepo
• other containers that run on this VPS under traefik have no issues
• traefik has no error logs

[noxify]

Had the same problem and found your post.

In our case, we have called an internal api endpoint via fetch to check the current session of an user ( we're using lucia-auth ).
We used request.nextUrl.origin initially to get the base url.
This worked fine locally, because the origin is/was localhost.

While deploying the app into our OpenShift cluster, the origin was the internal pod address.
This produces the mentioned error.

We solved it by setting an environment variable with the fallback to the origin.

 const origin = process.env.APPLICATION_URL
      ? // eslint-disable-next-line no-restricted-properties
        `https://${process.env.APPLICATION_URL}`
      : request.nextUrl.origin

[mr-chandan]

any solution ??

[mr-chandan]

i was using request.nextUrl.origin in my middleware.ts file i replaced it to http://127.0.0.1:3000 and it worked fine and also using http://localhost:3000 would not work

[hanssl]

same issue. no solution yet?

[fernandowski]

for everyone who is having this issue. I do not think it is a bug that needs to be fixed by vercel what is happening you are likely making an api call using an unsupported protocol in your middleware.

you can do something like this to figure out the protocol and host

const host = request.headers.get("host");
const isLocalHost = host?.startsWith("localhost") || host?.startsWith("127.0.0.1");
const protocol = isLocalHost ? "http" : "https";